the-decoder

China eyes export curbs on its top AI models, and Europe is caught in the middle

China is looking into restricting foreign access to its most advanced AI models. Authorities have already held talks with tech giants like Alibaba and Bytedance. For Europe, which relies heavily on ...
the-decoder

Tencent releases Hy3 open-source model that allegedly matches models up to five times its active size

Tencent has officially released its AI model Hy3. The model uses a Mixture-of-Experts (MoE) architecture with 295 billion total parameters. Of those, 21 billion are active at any given time, plus 3.8 ...
the-decoder

Meta follows SpaceX's playbook and builds a cloud business to sell its spare AI compute to outside customers

Meta is building its own cloud business to sell spare AI compute to outside customers. With planned AI investments of up to $145 billion this year alone, the same question that came up with xAI now ...
the-decoder

Half of Claude users say AI can already handle half their work according to Anthropic survey

About half of Claude users say AI can already handle half or more of their work. That's according to a survey of roughly 9,700 Claude users (Chat, Cowork, Code) by Anthropic. About 33 percent see AI ...
the-decoder

OpenAI's new flagship model GPT-5.6 Sol cheats on software tests more than any model before it

OpenAI's GPT-5.6 cheats a lot. That's the key finding from an independent evaluation by METR. During testing with software tasks, OpenAI's new flagship model GPT-5.6 Sol showed the highest rate of ...
the-decoder

Anthropic gets US approval to bring back Claude Mythos 5

The US government has approved Anthropic to redeploy Claude Mythos 5, according to the company. The model is Anthropic's most powerful AI model for cybersecurity and can now be used again by US ...
the-decoder

NYU finance professor Damodaran warns an AI crash could hit harder than the dot-com bust

Aswath Damodaran, a finance professor at New York University, warns that a potential crash in the AI sector could be more painful than the bursting of the dot-com bubble around 2000. In the podcast ...
the-decoder

OpenAI's Codex can now watch you work once and repeat the task forever

Also new in version 26.616: bulk actions for the Automations history, plus the ability to hand off threads between a local and remote host to continue tasks on a connected machine. Codex is OpenAI's ...
the-decoder

Microsoft researcher builds a working neural network out of goats in Age of Empires II to critique AI science

Adrian de Wynter, a researcher at Microsoft and the University of York, has built a working neural network inside the map editor of the legendary strategy game Age of Empires II. It sounds like a joke ...
the-decoder

Amazon, Nvidia, and AMD bet $310 million on AI startup building 3D world models

Amazon, Nvidia's and AMD's venture arms are investing $310 million in AI startup Odyssey ML, which builds world models that simulate the physical world in 3D. The round values Odyssey at $1.45 billion ...
the-decoder

Amazon and five other companies reportedly triggered the government crackdown on Anthropic's Fable model

Amazon CEO Andy Jassy and other tech executives warned the Trump administration about security risks in Anthropic's AI model, Fable, prompting swift government action. After Anthropic refused to pull ...
the-decoder

Microsoft's SkillOpt boosts GPT-5.5 by using nothing but a trained Markdown file

A simple Markdown file is apparently enough to boost GPT-5.5 by more than 20 points on procedural tasks. That's the promise of SkillOpt, a method from Microsoft and three Chinese universities that ...